---
title: "Codebook"
output: pdf_document
date: "2025-04-18"
---

## Introduction

This document provides information about the variables included in the original experimental datasets used in this paper.


### all_raw_data.rds

This dataset contains raw data from the three survey waves. The dataset is anonymized. It contains the following variables. The script `00_process_data.R` uses this dataset to produce `all_processed_data.rds` which is used in all of the analyses. 

- `response_id`: qualtrics generated id

- `q_race`: respondents' race (categorical)

- `income`: respondents' self-reported income level (categorical)

- `q_sub_wb_1`: item 1 (happiness) for subjective well-being  

- `q_sub_wb_2`: item 2 (depression) for subjective well-being  

- `q_sub_wb_3`: item 3 (anxiety) for subjective well-being  

- `q_sub_wb_4`: item 4 (isolated) for subjective well-being  

- `q_sub_wb_5`: item 5 (life satisfaction) for subjective well-being  

- `q_digital_lit_1`: item 1 in the digital literacy composite measure (7-point likert scale from strongly disagree to strongly agree)

- `q_digital_lit_2`: item 2 in the digital literacy composite measure (7-point likert scale from strongly disagree to strongly agree)

- `q_digital_lit_3`: item 3 in the digital literacy composite measure (7-point likert scale from strongly disagree to strongly agree)

- `q_digital_lit_4`: item 4 in the digital literacy composite measure (7-point likert scale from strongly disagree to strongly agree)

- `f_headline_1_acc`: accuracy question for headline 1 (false) 

- `f_headline_1_exp`: recall question for headline 1 (false)

- `f_headline_2_acc`: accuracy question for headline 2 (false)   

- `f_headline_2_exp`: recall question for headline 2 (false)

- `f_headline_3_acc`: accuracy question for headline 3 (false)   

- `f_headline_3_exp`:  recall question for headline 3 (false)

- `t_headline_4_acc`: accuracy question for headline 4 (false)   

- `t_headline_4_exp`: recall question for headline 4 (false)

- `t_headline_5_acc`: accuracy question for headline 5 (true)   

- `t_headline_5_exp`:  recall question for headline 5 (true) 

- `t_headline_6_acc`: accuracy question for headline 6 (true)   

- `t_headline_6_exp`: recall question for headline 6 (true)

- `t_headline_7_acc`: accuracy question for headline 7 (true)  

- `t_headline_7_exp`: recall question for headline 7 (true)

- `t_headline_8_acc`: accuracy question for headline 8 (true)  

- `t_headline_8_exp`: recall question for headline 8 (true)  

- `q_confidence`: confidence in headline accuracy responses (categorical) 

- `q_runoff`: intention to vote in the runoff election 

- `positive_part`: which party participant has positive feelings towards (categorical)  

- `negative_part`: which party participant has negative feelings towards (categorical)   

- `q_trust_out_1`: self-reported trust towards federal government (7-point  scale from low to high level of trust)  

- `q_trust_out_2`: self-reported trust towards Congress (7-point  scale from low to high level of trust)    

- `q_trust_out_3`: self-reported trust towards election authorities  (7-point  scale from low to high level of trust)    

- `q_trust_out_4` : self-reported trust towards TV Globo (7-point  scale from low to high levels of trust)   

- `q_trust_out_5`: self-reported trust towards National Newspapers (7-point  scale from low to high levels of trust)   

- `q_ideology_out_1`: self-reported ideological position (ordered from left to right)

- `q_ideology_out_7`: perceived ideological position of Lula (ordered from left to right) 

- `q_ideology_out_8`: perceived ideological position of Bolsonaro (ordered from left to right)

- `q_legitimacy`: trust in the electronic voting system in Brazil (categorial)  

- `affective_pol_1`: feelings scale towards Lula (7-point scale ordered from dislike to like) 

- `affective_pol_2`: feelings scale towards Bolsonaro (7-point scale ordered from dislike to like)  

- `affective_pol_3`: feelings scale towards Evangelicals (7-point scale ordered from dislike to like)  

- `affective_pol_4`: feelings scale towards LGBTQ people (7-point scale ordered from dislike to like)  

- `q_social_pol_1`: willingness to engage with an outgroup voter;  item being friends 

- `q_social_pol_2`: willingness to engage with an outgroup voter;  item dating  

- `q_social_pol_3`: willingness to engage with an outgroup voter; item watching a soccer game

- `q_social_pol_4`: willingness to engage with an outgroup voter; item their child dating 

- `q_social_pol_5`: willingness to engage with an outgroup voter; item spend Christmas

- `q_social_pol_6`: willingness to engage with an outgroup voter; item be in the same WhatsApp groups 

- `policy_sup_1`: policy agreement from low to high item 1 

- `policy_sup_2`: policy agreement from low to high item 2  

- `policy_sup_3`: policy agreement from low to high item 3  

- `policy_sup_4`: policy agreement from low to high item 4  

- `policy_sup_5` : policy agreement from low to high item 5 

- `q_news_source_1`:  substitution: how often participants used other social media platforms during the three weeks of the study (categorical)

- `q_news_source_2`: substitution: how often participants read news on the internet during the three weeks of the study (categorical)

- `q_news_source_3`: substitution: how often participants watched TV  during the three weeks of the study (categorical) 

- `q_news_source_4`: substitution: how often participants spent time with friends during the three weeks of the study (categorical)  

- `q_news_source_5` : substitution: how often participants spent time on other non-smartphone activities during the three weeks of the study (categorical)

- `q_social_media_usage_1`: substitution: WhatsApp (Categorical)

- `q_social_media_usage_2`: substitution: Facebook (Categorical)  

- `q_social_media_usage_3`: substitution: Twitter (Categorical)  

- `q_social_media_usage_4`: substitution: TikTok (Categorical)  

- `q_social_media_usage_5`: substitution: Telegram (Categorical)  

- `q_social_media_usage_7`: substitution: Youtube (Categorical)  

- `q_social_media_usage_8`: substitution: Instagram (Categorical)  

- `q_social_media_usage_9`: substitution: None (Categorical)  

- `q_whatsapp`: self-reported time spent on WhatsApp during the weeks of the study 

- `q_fakenews`: self-reported exposure to misinformation on WhatsApp during the weeks of the study 

- `q_compliance_1`: which type of media they guess they were not accessing during the weeks  of the study, option: about sports; only treatment  

- `q_compliance_4`: which type of media they guess they were not accessing during the weeks 
of the study, option: politics and elections; only treatment  

- `q_compliance_5`: which type of media they guess they were not accessing  during the weeks of the study, option: family; only treatment 

- `q_compliance_6`: which type of media they guess they were not accessing  during the weeks of the study, option: memes and jokes; only treatment 

- `q_compliance_7`: which type of media they guess they were not accessing  during the weeks of the study, option: work; only treatment 

- `q_compliance_direct`: self-reported compliance question

- `q_compliance_control_1`:  which type of media  they accessed during the weeks of the study, option: about sports; only control 

- `q_compliance_control_4`  which type of media  they accessed during the weeks of the study, option: politics and elections; only control

- `q_compliance_control_5`   which type of media  they accessed during the weeks of the study, option: family; only control

- `q_compliance_control_6`  which type of media they accessed to during the weeks of the study, option: memes and jokes; only control

- `q_compliance_control_7`  which type of media they guess  they accessed the weeks of the study, option: work; only control

- `q_com_direct_control`:    self-reported compliance measure only for the three days the control deactivated

- `q_spillover`: spillover measure for other WhatsApp usage 

- `q_device`: participant mobile device type

- `exp`: treatment assignment 

- `duration_in_min`: duration of the survey in minutes 

- `w1_q_age`: participants' pre-treatment age (numeric)

- `w1_q_gender`: participants' pre-treatment gender 

- `w1_q_education`: participants' pre-treatment education level

- `w1_q_politics`: participants' pre-treatment interest in politics 

- `w1_ideology_you`:  participants' pre-treatment ideological placement

- `w1_ideology_lula`:  pre-treatment perceived ideological position of Lula (ordered from left 
to right)

- `w1_ideology_bolsonaro`:  pre-treatment perceived ideological position of Bolsonaro (ordered from left to right)

- `w1_q_legitimacy`: self-reported trust in the election results  

- `w1_q_social_media`: social media platforms that participant has an account on

- `w1_q_whatsapp` self-reported WhatsApp usage  

- `w1_q_whatsapp_usage`: which device participant uses WhatsApp the most on  

- `w1_q_whatsapp_image`: pre-treatment WhatsApp exposure to politics through multimedia content 

- `w1_q_whatsapp_purposes_family`:  pre-treatment WhatsApp daily usage for communication with 
family

- `w1_q_whatsapp_purposes_pay_bills`  pre-treatment WhatsApp daily usage for paying bills

- `w1_q_whatsapp_purposes_news`  :  pre-treatment WhatsApp daily usage for news

- `w1_q_whatsapp_purposes_politics`: pre-treatment WhatsApp daily usage for politics  

- `w1_q_whatsapp_purposes_groups`: pre-treatment WhatsApp daily usage via large groups

- `w1_q_study`: consent to join the study

- `w1_whatsapp`: binary indicator for social media app usage - WhatsApp

- `w1_instagram`: binary indicator for social media app usage - Instagram

- `w1_tiktok`: binary indicator for social media app usage - TikTok 

- `w1_facebook`: binary indicator for social media  app usage - Facebook  

- `w1_telegram`: binary indicator for social media  app usage - Telegram  

- `w1_twitter`: binary indicator for social media  app usage - Twitter 

- `w1_other_social_media`: binary indicator for social  media app usage - others  

- `compliance_I`: difference in bytes consumption (in MB) between the first and last compliance screenshot

- `compliance_one_side`: binary measure of compliance as described in the main manuscript

- `w1_q_trust_government`: pre-treatment self-reported trust towards federal government (7-point scale from low to high level of trust)  

- `w1_q_trust_congress`: pre-treatment self-reported trust towards the congress (7-point  scale from low to high level of trust) 

- `w1_q_trust_local_government`: pre-treatment self-reported trust towards local governments (7-point  scale from low to high level of trust)  

- `w1_q_trust_supreme_court`: pre-treatment self-reported trust towards the supreme court (7-point  scale from low to high level of trust) 

- `w1_q_trust_electoral_authorities`: pre-treatment self-reported trust towards electoral authorities (7-point  scale from low to high level of trust) 

- `w1_q_trust_globo`  : pre-treatment self-reported trust towards TV Globo (7-point  scale from low to high level of trust) 

- `w1_q_trust_news_channels`  : pre-treatment self-reported trust towards News Channels (7-point  scale from low to high level of trust) 

- `w1_affective_pol_lula`: pre-treatment feelings scale towards Lula (7-point scale ordered from dislike to like)

- `w1_affective_pol_bolsonaro`: pre-treatment feelings scale towards Bolsonaro (7-point scale ordered from dislike to like)  

- `w1_affective_pol_lgbt`: pre-treatment feelings scale towards LGBT people (7-point scale ordered from dislike to like) 

- `w1_affective_pol_evangelicals`: pre-treatment feelings scale towards Evangelicals (7-point scale ordered from dislike to like) 

- `w1_affective_pol_catolics`: pre-treatment feelings scale towards Catolics (7-point scale ordered from dislike to like) 

- `w1_affective_pol_journalists` : pre-treatment feelings scale towards journalists (7-point scale ordered from dislike to like)

- `w1_affective_pol_electoral_authorities`: pre-treatment feelings scale towards Electoral Authorities (7-point scale ordered from dislike to like) 

- `w1_vote_first`: self-reported vote choice in the first round 

- `w1_q_fakenews`: pre-treatment self-reported exposure to misinformation on WhatsApp 

- `w1_q_device_treatment`: device type (Android vs iPhone) for treatment 

- `w1_q_device_control`: device type (Android vs iPhone) for control 

- `w1_exp`: experimental condition 

- `w1_trust_zcore`: z-score with all trust measures recoded 

- `w1_vote_first_recode`: recoded variable for vote choice in the first round 

- `w1_vote_left`: pre-treatment indicator if voting for the left 

- `w1_vote_right`: pre-treatment indicator  if voting for the right

- `w1_partisans`: pre-treatment indicator if voting null  

- `w1_affective_pol_diff_all`: pre-treatment absolute difference between feelings towards Lula and Bolsolnaro

- `w1_affective_social_diff_all` : pre-treatment absolute difference between feelings towards Evangelicals and LGBT people

- `w1_affect_pol_diff_partisans` : pre-treatment absolute difference between feelings towards Lula and Bolsolnaro only for partisans 

- `w1_affect_pol_social_partisans`: pre-treatment absolute difference between feelings towards Evangelicals and LGBT people only for partisans  

### compliance.csv

This dataset sumarises anonymized compliance status of all participants in the study. It contains the following variables:

- `exp`: experimental condition

- `compliance_I`: difference in bytes consumption (in MB) between the first and last compliance screenshot

### pre_treatment.rds

This dataset contains anonymized pre-treatment covariates used in the manuscript,  collected in waves 1 and waves 2 of the study (screening and treatment). It contains the following variables:

- `response_id`: treatment assignment survey Qualtrics ID for every respondent

- `exp`: treatment condition

- `invited_enrolled`: variable indicating participant was invited and enrolled vs invited but not enrolled               

- `invited_started`:  variable indicating participant was enrolled vs invited and at least 
started the treatment assignment survey vs those who never started the treatment assignment

- `invited_completed`: variable indicating participant completed the study vs those who dropped out in the post-treatment survey

- `enrolled`: variable indicating participant enrolled in the study vs not enrolled but who at least started the study      

- `completes`: indicator for completes  

- `Age`: participants' pre-treatment age (numeric)

- `Gender`: participants' pre-treatment gender (numeric) 

- `Education`: participants' pre-treatment education level  (numeric)  

- `Interest in Politics` participants' pre-treatment interest in politics (numeric)

- `Ideology Self`: participants' pre-treatment ideology (numeric)

- `Polarization Candidates`: difference in ideological position between Lula and Bolsonaro (numeric)  

- `WP:Daily time` : pre-treatment WhatsApp daily usage (numeric)          

- `WP:Frequency Politics`:   pre-treatment WhatsApp daily usage for politics (numeric)           
- `WP:Frequency News`: pre-treatment WhatsApp daily usage for news (numeric)  

- `WP:Frequency Family`: pre-treatment WhatsApp daily usage for communication with family (numeric)     

- `WP:Frequency Images about Politics`: pre-treatment WhatsApp exposure to politics through multimedia content (numeric)  

- `N Active Social Media Apps`: pre-treatment number of active social media accounts

- `Duration Pre-treatment`: duration of the screening survey    

- `q_fake_news` : self-reported exposure to misinformation on WhatsApp                      

- `affective_pol_diff_all`:  difference in feelings between Lula and Bolsonaro (numeric)     

- `Trust elections`: self-reported trust in the election results (numeric)     

- `location_latitude`: participant latitude extract from Qualtrics                 

- `location_longitude`: participant longitude extract from Qualtrics     

- `vote_first_recode`: participant pre-treatment and pre-election vote choice
